Case Study: Time Warner Cable's Formula for Maximizing Adobe Experience Manager

AEM Maxed! = Agile + Automation
Overcome three common challenges with AEM delivery
2015
Adobe
Summit

Cat
Reusswig,
Senior
Director,
Web
Portals
Engineering,
Time
Warner
Cable

Mark
Kelley,
Client
Services,
iCiDIGITAL

Presentation Agenda
1. Overview three common AEM delivery challenges
2. See how Time Warner Cable has overcome each of these
challenges and evolved over the years
3. Revealing of three open-source tools for AEM being
developed by TWC and iCiDIGITAL
4. Q&A
Questions? #ici #adobesummit

Presentation Contests
First to Answer the
Question Wins.
Ask a Question on Twitter
#ici #adobesummit
Most Favorites by 12PM,
Thursday, wins.

Common Challenges
When Implementing AEM
1. Lack of collaboration among business, IT, and marketing
stakeholders leads to inefficiencies and possibly failure
2. Creating and scaling the AEM delivery team to serve multiple
websites and business units
3. Inadequate automation & testing focus results in poor code
quality and site experience

Common AEM Challenge #1
Lack of collaboration among business, IT, and
marketing stakeholders leads to inefficiencies

CULTURE IS KEY
A way of thinking, behaving or working
that exists in an organization

2012
•  1 scrum team grew to 3 teams
•  Definition of done
•  Tedious manual deploy model
•  Agile infancy, esp on business side
2013
•  3 scrum teams grew to 6 teams
•  Improved automation testing
•  Maturing agile
•  Improved continuous integration
•  Improved user story development
2014
Remained at 6 teams for
major release Humming
along…
TWC Maturation over Three Years

TWC Definition of Done for Development:
l  Meets acceptance criteria
l  Is deployable
l  Been tested & code reviewed
Expanded later to:
l  Test in both author & publish and multi
Devices/Browsers
l  Generally code review & test review are +2
A Philosophy We Embrace

The model we began with in the early days
Development
Cycle(s)
Wires / CSS
Comps User Stories
CQ5
Components
Enhancements
Testing Dev
Automation Acceptance
OS /
Browsers Bug Fixes
Production
Content
Authoring QA Updates
Roll-out Strategy Prod Support Prioritization
Future
Phases
TWC Responsibility Model (2012)
Business
Responsibility
Development
Responsibility
Troubleshooting
Shared
Responsibility

TWC Responsibility Model (Today)
Business
Development
Intake
E-governance
UAT
Post Production Validation
Story/Defect Prioritization
Both
Story Review
Requirements/Design (Create Stories)
Backlog Management
eCommerce Priority Review
Sprint Planning
Mid-Sprint Demo
Sprint Closure
Daily Stand-up
Scrum of Scrums
Tasking
Dev / Test / CI
Production Release
Sprint Retro
Beta Release
Architecture
How and Why
DOTs Handoff
UAT Handoff
Go/No-Go
2-Week Sprint

– External and internal requests use Podio Request Form
- New requests are reviewed weekly
-  Request includes
l  Categories (for routing request to appropriate team)
l  Description of effort
l  Justification for effort
l  High level deliverables and requirements
l  Requested release dates
-  Request processes
l  Requests for content authoring directed to Authoring Team
(no development necessary).
l  Requests for new functionality, features or projects are directed
to eGovernance team for review and prioritization.
l  Project status and details are tracked within Podio ticket.
Product Request Intake Process & Routing

MVP – Minimal Viable Product, the smallest useful set of
capabilities that will meet the business need in production
Minimum Viable Product Product Vision
This came later too – and is a critical success factor now
MVP and Release Planning

P Purpose
O Outcomes
W What’s in it for me
E Expectations from
participants
R Result
Understand the business problem to be addressed
Inform product backlog
ROM Difficulty (Does story need to be split. Does it roughly match
business expectations of difficulty)
Test the edges of the solution recommended: RWD, failure conditions, failure
responses, success responses, do we have a spork that would fit the need, 508
considerations, patterns for reuse
Provide a readiness evaluation to the Product Manager, UX, Author, Dev
Team, Testing, and Management so everyone knows what we’re building and
why
•  Subset of Development and Testing Team
•  Subset of Authoring Team
•  Technical Product Owners (TPOs)
•  Product Owners
Attendees:
Story Review (POWER)

1. Held weekly
2. Incoming stories
l  Reviewed for completeness
l  Multi-Story efforts categorized and ranked into Epic
l  Prioritized per eGovernance priority list
l  Prioritized agenda set for Story Review
3. Defect Review
l  Reviewed for duplication
l  Reviewed for reproducibility
l  Confirmed it is an Omega defect
-  If not, directed to proper team for resolution
l  Reviewed to determine there is enough detail
For the Team to work including repo steps,
Screen shots, browser version, environment, etc.
5. Cherry-Picks
l  Requested for gating or urgent issues.
Ranking, Categorization, and Review of Stories and Defects
Story Defect / Prioritization
4. Rank order prioritized within categorical groups:
l  Gating Defects (P1) – Defects which will prevent a Sprint release
l  Critical Defects (P1.5) – Defects that are critical, but will not hold
up a Sprint release
l  Prioritized Defects (P2) – Defects which are considered high
priority
l  Backlogged Defects (P3) – Defects which can remedied as
resource availability allows
l  TechPriority / TechDebt Defects - Architectural or structural defects
for prioritization by technical team =20%

Creating and scaling the AEM delivery team
to serve multiple websites and business units

Shake up the teams
~every 6 months forges
greater strength
& relationships
6 Scrum Teams
l  Volunteer Scrum Master with Double Duty
(Dev/QA)
l  Team commits to stories accepted &
pointed
l  Fibonacci pointing
l  Team Norms, responsible for DONE!
l  Generally 6 members per team
l  Embedded QA
l  Mixed geography & time zones
Team Alignment with Autonomy
Horizontal Team
l  TPO
l  DevOps
l  QA Automation
l  Analytics
l  Architecture Forum
l  How & Why Forum
Support
l  DOTS/DevOps =
Support across all environments

P Purpose
O Outcomes
E Expectations from
participants
R Result
Development team discuss and commit to work for sprint
Discussion of story details and questions beyond what’s in the ticket.
Re-calibrate views of complexity for roadmap. Identify missing information.
Clarity of skillset and team availability to inform selection of work
Discussion of whether to roll or complete stories from previous sprint.
Evaluate work critically. Ownership of expected work. Share learning from
story refinement. Document decisions.
Clear picture of expected deliveries
Attendees: Scrum Team, Product Owner, Technical Product Owners, Cross functional team
Sprint Planning (POWER)

- Break Up Assigned Stories
•  Dissect stories/epics
•  Create as many sub-tasks as needed
•  Create investigation tasks if needed
- Collaborate As Needed
•  Contact TPO/Business with questions
•  Pull in other teams that have SMEs
•  Brainstorm on possible solutions
- Assign Work
•  Everyone should have at least 1 task
•  Stories should be worked by priority
•  Unassigned tasks get picked up as
bandwidth becomes available
Sprint Tasking

P Purpose
O Outcomes
E Expectations from
participants
R Result
Expose blockers beyond individual team control
Coordination among teams
Share information across teams, received help for your groups needs, learn
from others.
Bring a summary of the blockers & risks from your SoS, as well as any
crossteam coordination needs not yet socialized
Informs risks to the sprint (business and technical)
SoS format is similar to a stand-up (short, focused on what is not yet known)
Daily Stand-Up / Scrum of Scrums (SoS)

P Purpose
O Outcomes
E Expectations from
participants
R Result
Avoid the problem of sprint closure being the first time Dev team gets feedback
from the business
Confirm we are on the right path. Mid-sprint SOS with business
Early preview of functionality for business. Discuss proposed solutions and
ideas to solve problems. Potentially identify an immediate refactor candidate
(up to team whether to digest in current sprint)
Development – Share in progress on stories
Business – Have questions answered
Alignment and understanding across business and technical teams, scrum
teams.
Mid-Sprint Demo

P Purpose
O Outcomes
E Expectations from
participants
R Result
Demo the agreed upon sprint work to the stakeholders
Inform progress against roadmap. Clarification on any possible
blockers or incomplete work.
Confirmation that our sprint work is “done”
Prepare to demo work and knowing sprint work is complete
The completion of the sprint
Sprint Closure

-  Reflect on previous sprint’s work
-  Each team executes as works best for them
-  Find out what was good, bad, or neutral
-  Approximately 1-2 hours (as needed)
-  Determine action items
-  Evaluate action items from previous sprint
-  Try to avoid repeating mistakes
-  EVERYONE PARTICIPATES
Sprint Retrospective

O Outcomes
E Expectations from
participants
R Result
Share technical knowledge among technical team, across teams Get a
common sense of best practices
Document via Wiki or web links
Learn something new from others’ experiences
Don’t have to make the same review comments repeatedly
Time is not wasted by presenting information that is common or not
useful. Presented topics are documented well enough to discuss.
Developers write better code and use tools more effectively
Attendees: Developers, QA, DevOps (Technical Folks)
P Purpose
How and Why

“Only thing we have to fear is fear itself” FDR
•  Don’t be afraid to take a big
step/bite – we added close to
10 people at a time
•  We failed a bunch – don’t be
afraid to try, fail, and adjust
•  Do retrospectives & actually
take action
•  Create horizontal forums
•  We reorganized and added
roles – TPO, Architect,
DevOps, Tier1, ++
•  The business team is
completely different & has a
different set of roles
•  We consistently do more face-
to-face whether virtual or live
Our Approach to the AEM Unknown

-  Onboarding ease
-  Horizontal consistency inducement
-  Peer mentorship
-  Team acceptance
-  Team working agreements
-  Training
Critical Success Factors to
Growing the Team

Inadequate automation & testing focus results
In poor code quality and site experience

Tools used to help automate the agile
process and develop test-driven culture

Deploy &
Validate Environment
Deploy &
Validate Environment
Integration Tests
Integration Tests
Build Unit Tests Sonar
Build Unit Tests Sonar
Functional Test (no UI)
Functional Test (no UI)
UI Tests
(FF, IE, & Chrome)
UI Tests
(FF, IE, & Chrome)
If successful
If successful
If failure, email
Everyone!
If failure, email
DevOps
If failure, email
Everyone!
If failure, email
Everyone!
Dev/Test/Cl (CI)
If failure, email
Everyone!

CR
+1
Testing+1
GHE = +1 creates a culture of/for code

CI kicks off with developer commits
Automated JIRA Comments
Testing Directed by Jenkins
SonarQube code quality
Performance with WebPageTest
Private instance of WebPageTest (WPT) server
All Chrome UITests automatically save performance results to WPT server
Visual breakdown of all page content load times
GHE = +1 creates a culture of/for code

GebConfig.groovy
‘firefox’ {
driver = “firefox”
}
‘ipad’ {
driver = “org.openqa.selenium.iphone.IPhoneDriver”
}
‘android-grid’ {
driver = { new RemoteWebDriver(new URL(seleniumGridUrl),
DesiredCapabilities.android()) }
‘ie8-winXP’ {
DesiredCapabilities caps = DesiredCapabilities.internetExplorer();
caps.setCapability(“platform”, “Windows 2003”);
caps.setCapability(“version”, “8”); caps.setCapability(“name”,
“regression-IE8-WinXP”)
driver = { new RemoteWebDriver(new URL(seleniumGridUrl), caps) }
}Specifying Tests with Spock
Technologies
Coding with Groovy
Organizing with Geb
Specifying with Spock
Directing with Jenkins
Running with Gradle
Managing with Grid2 and Vagrant
Diversifying with SaucelLabs
}Diversifying tests w SauceLabs
Windows VirtualBox images with
“vagrant up” attaches to 10 min
Local browser
Local emulator
Remote VMs
Attached to Grid
SauceLabs VM
Attached to Grid
31

Beta Release / UAT Handoff
- UAT Handoff
•  Review Wikis (UAT Team)
•  Meet with DOTS team (or representative)
•  Ask Questions
•  Prepare to Test
- Beta Release
•  Cut Code and Create Release Candidate
•  Deploy to Beta (Test) Environment
•  Mark tickets for handoff
•  Provide UAT Wikis
32

- UAT includes:
•  Functional GUI testing of planned release effort
•  Content and image verification
•  Regression testing of existing functionality
•  Responsive testing on mobile devices
•  Cross browser validation
•  Defect validation from external reporters
•  Test and Target
•  Analytic and GTM/DTM testing coordination
- New defects are logged for Story/Defect
Prioritization review
- Critical issues are tagged and escalated
for immediate review by product team
User Acceptance Testing (UAT)
33

- Represented Teams
•  Residential
•  Checkout
•  Offer Configuration
•  Authoring
•  UAT
•  Analytics
•  GTM / DTM
•  Test & Target
Are we all on-board?
Go / No-Go
- Final release sign-off from all impacted
teams
- Review of UAT status
•  Confirm all tickets tested and status
•  Review impact of any defects being released
•  Confirm all gating issues resolved
34

What is AppDynamics?
An application performance management software
What does it for?
Designed to help dev and ops troubleshoots problems in
complex production app
Key Benefits
-Get complete visibility into application performance
-Monitoring business Transaction in real time
-Set up proactive alerting to find problems before they
Hurt your business
-Fix problem faster than ever before.
Any build to any environment
Relic / Artifactory
Artifactory is a Content Reponsitory which, uses the Apache Jackrabbit
implementation. This natural compatibility makes it extremely easy to manage
fully-indexed and searchable metadata
Versions stored in artifactory
Automatic Confluence updates – which code is where
Push Button Environment Deploys
DevOps
-Ever try to sift through log files across multiple servers trying to find an
individual error or a trend of errors?
-Splunk presents all your data from multiple sources and makes it available
from a single location, making it easy to search for patterns.
-Some of the things that we have leveraged splunk for:
l  We Detecting Dos attack occurring on our rr.com site
l  Used it for various analytics to learn about our customers
l  Used if for capacity planning and finding load balancing problems.
l  We have identified code defects by trending errors
l  and many many more...
Puppet is IT automation software that help system
administrators manage infrastructure throughout its
lifecycle, from provisioning and configuration to
orchestration and reporting. Using Puppet, you can
easily automate repetitive tasks, quickly deploy critical
applications, and proactively manage changes,
scaling from 10s of servers to 1000s, on-premise
or in the cloud.
35

Production Release / DOTS Handoff
- DOTS Handoff
•  Meet with New DOTS teams
•  Relay any issues that are unresolved
•  Note any Gotchas or Pitfalls
•  Update Wiki
- Production Release
•  Cut Code
•  Prepare Beta Environments for Test
•  Handle Beta and Prod Issues
•  Deploy Release
•  Provide DOTS Handoff
36

Did it work as expected?
Post Production Validation
- Validating tickets includes
•  Tickets included in the release are working as expected
•  Mini regression confirming core functionality is working as expected (not a full
regression suite)
•  Mini browser compatibility and device testing suite
•  Validation of production content per copy decks (testing environment uses non-
confirmed contents)
•  Offer validation
- New defects are logged for Story / Defect Prioritization review
- Critical issues are tagged and escalated for immediate review by
production team
37

38
Some Coming Open Source Attractions J

Grabbit
Provide a fast and reliable way of copying content from
one Sling (specifically Adobe CQ / AEM) instance to
another.
https://github.com/TWCable/grabbit

CQ Gradle Plug-ins
A number of plugins to make it easer to work with
Adobe AEM from Gradle (Gradle.org), including CQ
Package, SCR, Sling Bundle, and Bundlor.
https://github.com/TWCable/cq-gradle-plugin

Jackalope
An in-memory implementation of the JCR with stubbing
capabilities for Apache Sling, better enabling unit
testing of JCR representations with complex objects
and structures.
https://github.com/TWCable/jackalope

Questions
2015 Adobe Summit
Cat Reusswig, TWC & Mark Kelley, iCiDIGITAL

Case Study: Time Warner Cable's Formula for Maximizing Adobe Experience Manager

More Related Content

Case Study: Time Warner Cable's Formula for Maximizing Adobe Experience Manager